Resource failure and buffer space allocation control for automated manufacturing systems

Abstract

Blockage and deadlock are undesirable in an automated manufacturing system (AMS). Thus, addressing these issues is essential to ensure smooth production evolution via system supervisory control. If an AMS contains unreliable resources which may break down unexpectedly, then the situation is complicated and problems become challenging. This work investigates the robust supervision for such AMSs from the perspective of management and control of resource allocation. The ultimate goal of this study is to develop a controller for allocating buffer space, such that any failure does not propagate through blocking to stall the rest of the system. In other words, the continual production of every part type not requiring any of the failed resources must be guaranteed at all time. We propose a supervisory control policy that satisfies the above desired property by using formal tools of automata and Petri nets. The policy relaxes the assumption made in previous work that each part type requires at most one failure-prone resource in the predefined production route. Meanwhile, the policy execution and implementation do not need any central buffers. We also present a survey and some comparative studies of various approaches in the literature about robust supervision for deadlock resolution in AMSs.